Method and device for determining decay of artificial intelligence model in wireless communication system

Abstract

The present disclosure relates to a 5G or 6G communication system for supporting a higher data transmission rate than a 4G communication system such as LTE. A terminal, according to one embodiment, may operate in a sequence of: transmitting, to a base station, channel inference capability information including information about an artificial intelligence model used for channel inference of the terminal; receiving, from the base station, configuration information about at least one CSI-RS used for channel inference, on the basis of the channel inference capability information; inferring at least one channel by using the artificial intelligence model on the basis of the configuration information; and determining decay of the artificial intelligence model on the basis of the inferred at least one channel.

Description

Method and device for determining deterioration of an artificial intelligence model in a wireless communication system

[0001] The present disclosure relates to the operation of a terminal and a base station in a wireless communication system. Specifically, the present disclosure relates to a device and method for determining the deterioration of an artificial intelligence model based on channel inference using an artificial intelligence model.

[0002] Looking back at the evolution of wireless communication over successive generations, technologies have primarily been developed for human-facing services such as voice, multimedia, and data. With the commercialization of 5G (5th-generation) communication systems, an explosive increase in connected devices is expected to be connected to communication networks. Examples of networked objects include vehicles, robots, drones, home appliances, displays, smart sensors installed in various infrastructures, construction equipment, and factory equipment. Mobile devices are expected to evolve into diverse form factors, including augmented reality glasses, virtual reality headsets, and holographic devices. In the 6th-generation (6G) era, efforts are being made to develop improved 6G communication systems to connect hundreds of billions of devices and objects and provide diverse services. For this reason, 6G communication systems are often referred to as "beyond 5G."

[0003] The 6G communication system, expected to be realized around 2030, will have a maximum transmission speed of terabytes per second (i.e., 1,000 gigabits per second) and a wireless latency of 100 microseconds (μsec). In other words, compared to 5G, the transmission speed in a 6G communication system will be 50 times faster, while the wireless latency will be reduced to one-tenth.

[0004] To achieve these high data rates and ultra-low latency, 6G communication systems are being considered for implementation in the terahertz band (e.g., from 95 gigahertz (GHz) to 3 terahertz (THz)). Compared to the millimeter wave (mmWave) band introduced in 5G, the terahertz band is expected to experience more severe path loss and atmospheric absorption, making it more crucial to ensure signal reach, or coverage, in this band. Key technologies to ensure coverage include radio frequency (RF) components, antennas, new waveforms that offer better coverage than OFDM (orthogonal frequency division multiplexing), beamforming, and multiple antenna transmission technologies such as massive multiple-input and multiple-output (MIMO), full-dimensional MIMO (FD-MIMO), array antennas, and large-scale antennas. In addition, new technologies such as metamaterial-based lenses and antennas, high-dimensional spatial multiplexing using orbital angular momentum (OAM), and reconfigurable intelligent surfaces (RIS) are being discussed to improve the coverage of terahertz band signals.

[0005] In addition, in order to improve frequency efficiency and system network, 6G communication systems are developing full duplex technology that utilizes the same frequency resources for uplink and downlink at the same time; network technology that integrates satellites and high-altitude platform stations (HAPS); network structure innovation technology that supports mobile base stations and enables optimization and automation of network operation; dynamic spectrum sharing technology through collision avoidance based on spectrum usage prediction; AI-based communication technology that utilizes artificial intelligence (AI) from the design stage and internalizes end-to-end AI support functions to realize system optimization; and next-generation distributed computing technology that realizes services with complexity that exceeds the limits of terminal computing capabilities by utilizing ultra-high-performance communication and computing resources (mobile edge computing (MEC), cloud, etc.). In addition, efforts are being made to further strengthen connectivity between devices, further optimize networks, promote softwareization of network entities, and increase the openness of wireless communications through the design of new protocols to be used in 6G communication systems, the implementation of hardware-based security environments, the development of mechanisms for the safe use of data, and the development of technologies for maintaining privacy.

[0006] Research and development of these 6G communication systems are expected to enable a new level of hyper-connected experience through the hyper-connectivity of 6G communication systems, which encompass not only connections between things but also connections between people and things. Specifically, 6G communication systems are expected to enable services such as truly immersive extended reality (Truly Immersive XR), high-fidelity mobile holograms, and digital replicas. Furthermore, services such as remote surgery, industrial automation, and emergency response, which are provided through enhanced security and reliability, will find application in diverse fields such as industry, healthcare, automotive, and home appliances.

[0007] According to one embodiment, a terminal may include a transceiver and at least one processor connected to the transceiver. The at least one processor may transmit channel inference capability information including information regarding an artificial intelligence model used for channel inference of the terminal to a base station. The at least one processor may receive, from the base station, configuration information regarding at least one CSI-RS used for the channel inference based on the channel inference capability information. The at least one processor may infer at least one channel through the artificial intelligence model based on the configuration information. The at least one processor may determine degradation of the artificial intelligence model based on the at least one inferred channel.

[0008] A base station according to one embodiment may include a transceiver and at least one processor connected to the transceiver. The at least one processor may receive channel inference capability information from a terminal, the channel inference capability information including information about an artificial intelligence model used for channel inference of the terminal. The at least one processor may transmit configuration information regarding at least one CSI-RS used for channel inference based on the channel inference capability information.

[0009] A method of operating a terminal according to one embodiment may include a step of transmitting channel inference capability information including information regarding an artificial intelligence model used for channel inference of the terminal to a base station. A method of operating a terminal according to one embodiment may include a step of receiving, from the base station, configuration information regarding at least one CSI-RS used for the channel inference based on the channel inference capability information. A method of operating a terminal according to one embodiment may include a step of inferring at least one channel through the artificial intelligence model based on the configuration information. A method of operating a terminal according to one embodiment may include a step of determining decay of the artificial intelligence model based on the at least one inferred channel.

[0010] A method of operating a base station according to one embodiment may include a step of receiving, from a terminal, channel inference capability information including information regarding an artificial intelligence model used for channel inference of the terminal. A method of operating a base station according to one embodiment may include a step of transmitting, to the terminal, configuration information regarding at least one CSI-RS used for channel inference based on the channel inference capability information.

[0011] Figure 1 is a diagram illustrating the basic structure of the time-frequency domain, which is a wireless resource domain of a wireless communication system.

[0012] FIG. 2 is a diagram illustrating a frame, subframe, and slot structure in a wireless communication system.

[0013] FIG. 3 is a diagram illustrating an example of measuring a channel based on CSI-RS in the time domain.

[0014] FIG. 4 is a block diagram illustrating components of a terminal according to one embodiment.

[0015] FIG. 5 is a diagram for explaining an artificial intelligence model operated in a terminal according to one embodiment.

[0016] FIG. 6 is a diagram showing a terminal according to one embodiment that operates in conjunction with a server.

[0017] FIG. 7 is a block diagram illustrating components of a base station according to one embodiment.

[0018] Figure 8 is a flowchart illustrating a method of operating a terminal according to one embodiment.

[0019] Figure 9 is a flowchart illustrating a method of operating a terminal according to one embodiment.

[0020] Fig. 10 is a flowchart illustrating an operation method of a terminal according to one embodiment.

[0021] Fig. 11 is a flowchart illustrating an operation method of a base station according to one embodiment.

[0022] Fig. 12 is a flowchart illustrating an operation method of a base station according to one embodiment.

[0023] FIG. 13 is a flowchart illustrating an operation method of a terminal according to one embodiment and a base station according to one embodiment.

[0024] FIG. 14 is a flowchart illustrating an operation method of a terminal according to one embodiment and a base station according to one embodiment.

[0025] The terms used in this specification will be briefly explained, and the present invention will be described in detail.

[0026] The terms used in this invention have been selected from widely used, current terms, taking into account the functions of the invention. However, these terms may vary depending on the intentions of those skilled in the art, precedents, the emergence of new technologies, etc. Furthermore, in certain cases, terms may be arbitrarily selected by the applicant, in which case their meanings will be described in detail in the relevant description of the invention. Therefore, the terms used in this invention should not be defined simply as names, but rather based on their inherent meanings and the overall content of the invention.

[0027] When a part of the specification is said to "include" a component, this does not exclude other components, but rather implies the inclusion of other components, unless otherwise specifically stated. Furthermore, terms such as "part," "module," etc., used throughout the specification refer to a unit that processes at least one function or operation, which may be implemented in hardware, software, or a combination of hardware and software.

[0028] Additionally, the description 'at least one of A, B, and C' means that it can be any one of 'A', 'B', 'C', 'A and B', 'A and C', 'B and C', and 'A, B, and C'.

[0029] It should be understood that the combinations of blocks and sequence diagrams in each flowchart can be performed by one or more computer programs containing computer-executable instructions. The one or more computer programs may be stored entirely in a single memory or may be divided and stored in multiple different memories.

[0030] All functions or operations described in this document may be performed by a single processor or a combination of processors. A single processor or a combination of processors is a circuitry that performs processing, and may include circuitry such as an Application Processor (AP), a Communication Processor (CP), a Graphical Processing Unit (GPU), a Neural Processing Unit (NPU), a Microprocessor Unit (MPU), a System on Chip (SoC), or an Integrated Chip (IC).

[0031] A processor may include various processing circuits and / or multiple processors. For example, the term “processor” as used herein, including in the claims, may include various processing circuits, including at least one processor. At least one processor, one or more processors, may be individually and / or collectively configured to perform the various functions described herein in a distributed fashion. As used herein, “processor,” “at least one processor,” and “one or more processors” may be configured to perform multiple functions. However, these terms encompass, without limitation, situations where one processor performs some of the functions and other processor(s) perform other parts of the functions, and situations where a single processor may perform all of the functions. Furthermore, the at least one processor may include a combination of processors that perform various of the disclosed functions in a distributed manner. At least one processor may execute program instructions to achieve or perform various functions.

[0032] Below, with reference to the attached drawings, embodiments of the present invention are described in detail so that those skilled in the art can easily implement the present invention. However, the present invention can be implemented in various different forms and is not limited to the embodiments described herein. In the drawings, parts irrelevant to the description have been omitted to clearly explain the present invention, and similar parts have been designated with similar reference numerals throughout the specification.

[0033] In the embodiments of this specification, the term “user” may mean a person who controls a system, function, or operation.

[0034] Figure 1 is a diagram illustrating the basic structure of the time-frequency domain, which is a wireless resource domain of a wireless communication system.

[0035] In Fig. 1, the horizontal axis represents the time domain, and the vertical axis represents the frequency domain. The basic unit of resources in the time and frequency domains is a resource element (RE, 101), which can be defined as one OFDM (Orthogonal Frequency Division Multiplexing) symbol (or DFT-s-OFDM (discrete Fourier transform spread OFDM) symbol) (102) in the time axis and one subcarrier (subcarrier, 103) in the frequency axis. In the frequency domain (For example, 12) consecutive REs can form one resource block (RB, 104). Also, in the time domain, A number of consecutive OFDM symbols can constitute one subframe (subframe, 110).

[0036] FIG. 2 is a diagram illustrating a frame, subframe, and slot structure in a wireless communication system.

[0037] One frame (200) can be defined as 10ms. One subframe (201) can be defined as 1ms, and therefore one frame (200) can be composed of a total of 10 subframes (201). In addition, one slot (202, 203) can be defined as 14 OFDM symbols (i.e., the number of symbols per slot ( )=14). One subframe (201) may be composed of one or more slots (202, 203), and the number of slots (202, 203) per one subframe (201) may vary depending on μ(204, 205), which is a setting value for the subcarrier spacing.

[0038] In an example of FIG. 2, slot structures are illustrated for cases where the subcarrier spacing setting value is μ=0 (204) and μ=1 (205). When μ=0 (204), one subframe (201) can be composed of one slot (202), and when μ=1 (205), one subframe (201) can be composed of two slots (203). That is, depending on the setting value μ for the subcarrier spacing, the number of slots per subframe ( ) may vary, and accordingly the number of slots per frame ( ) may vary. Depending on the subcarrier spacing setting μ and can be defined as shown in Table 1 below.

[0039] [Table 1]

[0040]

[0041] FIG. 3 is a diagram illustrating an example of measuring a channel based on CSI-RS in the time domain.

[0042] In a wireless communication system, it is necessary to measure the status of the channel through which data is transmitted and received in order to ensure smooth data transmission and reception between the base station and the terminal.

[0043] The base station can transmit a CSI-RS (channel state information reference signal) (310) to the terminal in one slot per period to measure the state of the channel.

[0044] A base station may transmit a CSI-RS to a terminal in at least one symbol within a slot to measure the channel status. For example, one or more CSI-RSs may be transmitted in at least one symbol unit within a slot.

[0045] The terminal can measure the status (320) of the channel through which data is to be transmitted and received based on the CSI-RS (310) received from the base station.

[0046] However, if the base station needs to transmit multiple CSI-RSs (330, 350) to the terminal or a new cycle of CSI-RSs (330, 350) depending on the state of the measured channel, the base station needs to newly designate a slot in which the multiple CSI-RSs (330, 350) to be transmitted to the terminal or the new cycle of CSI-RSs (330, 350) are transmitted.

[0047] In this case, the base station needs to transmit additional information or a message to the terminal to indicate that a slot in which multiple CSI-RSs (330, 350) or a new cycle of CSI-RSs (330, 350) are transmitted has been newly designated.

[0048] The terminal can measure a plurality of CSI-RSs (330, 350) or a channel (340, 360) corresponding to a new cycle of CSI-RSs (330, 350) by receiving additional information or messages from the base station using an artificial intelligence model.

[0049] A terminal including an artificial intelligence model can train the artificial intelligence model based on information or messages received from a base station.

[0050] However, when the terminal trains the AI ​​model by receiving additional information or a message from the base station indicating that multiple CSI-RSs or a slot in which a new cycle of CSI-RSs is transmitted has been newly designated, degradation of the AI ​​model may occur as the distribution of input data changes.

[0051] Degradation can mean that the performance of an AI model is degraded.

[0052] The terminal needs to manage the degraded artificial intelligence model to correctly measure the channel status.

[0053] Managing a degraded artificial intelligence model may include at least one of replacing the degraded artificial intelligence model with a new type of artificial intelligence model or re-training the degraded artificial intelligence model.

[0054] Therefore, in order to manage the artificial intelligence model, the terminal needs to monitor whether the artificial intelligence model has deteriorated.

[0055] Hereinafter, a specific method for determining deterioration of an artificial intelligence model by having a terminal and / or a base station infer / measure the state of a channel through an artificial intelligence model is disclosed.

[0056] FIG. 4 is a block diagram illustrating components of a terminal according to one embodiment.

[0057] A terminal (400) according to one embodiment may include a processor (410) and a transceiver (420).

[0058] A terminal (400) according to one embodiment may further include a memory (not shown) that stores at least one command.

[0059] The memory may store various data, programs or applications for driving and controlling the terminal (400) according to one embodiment. The memory may include, for example, a non-volatile memory including at least one of a flash memory type, a hard disk type, a multimedia card micro type, a card type memory (e.g., SD or XD memory, etc.), a ROM (Read-Only Memory), and an EEPROM (Electrically Erasable Programmable Read-Only Memory), a PROM (Programmable Read-Only Memory), and a volatile memory such as a RAM (Random Access Memory) or an SRAM (Static Random Access Memory).

[0060] The memory may store instructions, data structures, and program codes that can be read by the processor (410). In the following embodiments, the processor (410) may be implemented by executing instructions or codes of a program stored in the memory.

[0061] The processor (410) is a configuration that controls a series of processes so that the terminal (400) operates according to the embodiments described below, and may be composed of one or more processors.

[0062] The processor (410) may be composed of hardware components that perform arithmetic, logic, and input / output operations and signal processing. One or more processors included in the processor (410) may be circuitry such as a System on Chip (SoC), an Integrated Circuit (IC), etc. The processor (410) may be composed of at least one of, for example, a Central Processing Unit (CPU), a microprocessor, a Graphic Processing Unit (GPU), Application Specific Integrated Circuits (ASICs), Digital Signal Processors (DSPs), Digital Signal Processing Devices (DSPDs), Programmable Logic Devices (PLDs), and Field Programmable Gate Arrays (FPGAs), but is not limited thereto.

[0063] The processor (410) can write data to memory, read data stored in memory, and process data according to predefined operation rules, particularly by executing a program or at least one instruction stored in memory.

[0064] The transceiver (420) can communicate with an external device or server through at least one wired or wireless communication network by means of the processor (410).

[0065] The transceiver (420) may include at least one short-range communication module that performs communication according to a communication standard such as Bluetooth, Wi-Fi, BLE (Bluetooth Low Energy), NFC / RFID, Wi-Fi Direct, UWB, or ZIGBEE, and a long-range communication module that performs communication with a server to support long-range communication according to a long-range communication standard. The long-range communication module may perform communication through a communication network according to a 3G, 4G, 5G, and / or 6G communication standard, or a network for Internet communication.

[0066] According to one embodiment, a processor (410) may transmit channel inference information including information about an artificial intelligence model used for channel inference of a terminal (400) to a base station.

[0067] A processor (410) according to one embodiment can store at least one artificial intelligence model in memory.

[0068] A detailed description of the artificial intelligence model operated in the terminal (400) according to one embodiment will be described later in FIG. 5.

[0069] According to one embodiment, a processor (410) can receive at least one artificial intelligence model for which learning has been completed from a server through a transceiver (420).

[0070] A specific description of how a terminal (400) according to one embodiment receives at least one artificial intelligence model for which learning has been completed from a server will be described later in FIG. 6.

[0071] According to one embodiment, the processor (410) can store at least one artificial intelligence model for which learning has been completed, received from a server, in memory.

[0072] A processor (410) according to one embodiment can control an operation performed in a terminal (400) according to one embodiment through an artificial intelligence model.

[0073] Channel inference capability information may include at least one of the following: an identifier of the AI ​​model, the size of the AI ​​model, the complexity of the AI ​​model, information about the input of the AI ​​model, or information about the output of the AI ​​model, but is not limited thereto.

[0074] According to one embodiment, the processor (410) may assign an identifier to each of at least one artificial intelligence model.

[0075] According to one embodiment, the processor (410) may assign an identifier to each of at least one artificial intelligence model based on at least one of indexing or bitmap.

[0076] According to one embodiment, the processor (410) may assign an identifier to each of at least one artificial intelligence model according to the purpose of use of the at least one artificial intelligence model.

[0077] For example, if the first artificial intelligence model stored in the terminal (400) is an artificial intelligence model for beam management, the processor (410) may assign “identifier 1” to the first artificial intelligence model.

[0078] For example, if the second artificial intelligence model is an artificial intelligence model for positioning accuracy that accurately measures the location of the terminal, the processor (410) may assign “identifier 2” to the second artificial intelligence model.

[0079] For example, the processor (410) may assign “identifier 3” to the third artificial intelligence model if the third artificial intelligence model is an artificial intelligence model for channel inference.

[0080] According to one embodiment, the processor (410) can distinguish at least one artificial intelligence model based on at least one of the number of layers of the artificial intelligence model or learning data.

[0081] According to one embodiment, the processor (410) may assign an identifier to each of at least one artificial intelligence model included in at least one artificial intelligence model assigned an identifier according to the purpose of use.

[0082] For example, the processor (410) can distinguish at least one artificial intelligence model included in the third artificial intelligence model assigned “identifier 3”.

[0083] According to one embodiment, the processor (410) may assign an identifier to each of at least one artificial intelligence model included in the third artificial intelligence model.

[0084] According to one embodiment, the processor (410) may assign an identifier of “identifier 3, type A” to the artificial intelligence model A included in the third artificial intelligence model.

[0085] According to one embodiment, the processor (410) may assign an identifier of “identifier 3, type B” to the artificial intelligence model B included in the third artificial intelligence model.

[0086] However, this is merely an example, and the method by which the processor (410) assigns an identifier to at least one AI model according to one embodiment is not limited thereto. In another example, the identifier indicating that the third AI model is an AI model for channel inference and the identifier indicating the type of AI model used for channel inference may exist as separate parameters or fields.

[0087] However, this is only an example of a processor (410) according to one embodiment assigning an identifier to at least one artificial intelligence model, and is not limited thereto.

[0088] The size of the AI ​​model may include at least one parameter value, including the number of layers used for learning the AI ​​model or weights, but is not limited thereto.

[0089] The complexity of an AI model can refer to the amount of computation required to train the AI ​​model.

[0090] Information about the input of an artificial intelligence model can refer to data entered as input into the input stage of an artificial intelligence model.

[0091] For example, information about the input of an artificial intelligence model may include, but is not limited to, vector values ​​and / or matrix values ​​for at least one of the number of slots in the time domain or the interval between adjacent slots.

[0092] For example, information about the input of the artificial intelligence model may include, but is not limited to, vector values ​​and / or matrix values ​​for at least one of the number of resource blocks in the frequency domain or the spacing between adjacent resource blocks.

[0093] For example, information about the input of the artificial intelligence model may include, but is not limited to, vector values ​​and / or matrix values ​​for at least one of the number of sub-carriers in the frequency domain or the spacing between adjacent sub-carriers.

[0094] Information about the output results of an artificial intelligence model can mean the output according to the input of the artificial intelligence model.

[0095] For example, information regarding the output result of an artificial intelligence model may include information indicating that "the symbol of the second slot located four slots after the first slot is predictable", which is an output corresponding to the input, when the number of slots and / or the spacing between adjacent slots is input to the input terminal of the artificial intelligence model. However, this is not limited thereto.

[0096] For example, information regarding the output result of an artificial intelligence model may include information indicating that, when the number of slots and / or the spacing between adjacent slots is input to the input terminal of the artificial intelligence model, "the symbol of the second slot located 6 slots after the first slot and the symbol of the third slot located 4 slots after the second slot can be predicted," which is an output corresponding to the input. However, this is not limited thereto.

[0097] For example, information regarding the output result of an artificial intelligence model may include information indicating that "a symbol can be predicted in units of two resource blocks from the first resource block", which is an output corresponding to the input, when the number of resource blocks and / or the spacing between adjacent resource blocks is input to the input stage of the artificial intelligence model. However, this is not limited thereto.

[0098] For example, information about the output result of the artificial intelligence model may include information indicating that, when at least one of the number of slots and / or the spacing between adjacent slots or the number of resource blocks and / or the spacing between adjacent resource blocks is input as an input to the input terminal of the artificial intelligence model, "in the second slot located four slots after the first slot, a symbol can be predicted in units of two resource blocks from the first resource block," which is an output corresponding to the input. However, the present invention is not limited thereto.

[0099] According to one embodiment, a processor (410) may transmit channel inference capability information to a base station through a transceiver (420) based on at least one of uplink control information (UCI), medium access control-control element (MAC-CE), or radio resource control signaling (RRC signaling).

[0100] According to one embodiment, a processor (410) may receive configuration information regarding at least one CSI-RS used for channel inference from a base station based on channel inference capability information through a transceiver (420).

[0101] According to one embodiment, a processor (410) may receive configuration information regarding at least one CSI-RS from a base station based on at least one of downlink control information (DCI), MAC-CE, or RRC signaling.

[0102] The configuration information regarding at least one CSI-RS may include, but is not limited to, at least one of pattern information of at least one CSI-RS for inferring at least one channel or location information of at least one CSI-RS for inferring at least one channel.

[0103] The configuration information regarding at least one CSI-RS may be transmitted from the base station to another terminal in one embodiment in the form of an index and / or a codebook.

[0104] The pattern information of at least one CSI-RS may include periodic information of at least one CSI-RS transmitted from a base station to a terminal.

[0105] The periodic information of at least one CSI-RS may include at least one of periodic, semi-persistent, or aperiodic.

[0106] According to one embodiment, the processor (410) may receive activation information from the base station for determining whether to infer at least one channel through at least one of MAC-CE or DCI, if the pattern information indicates that the period of at least one CSI-RS for inferring at least one channel is semi-stationary.

[0107] A processor (410) according to one embodiment may receive trigger information from a base station via DCI that instructs to infer at least one channel, if the pattern information indicates that the period of at least one CSI-RS for inferring at least one channel is aperiodic.

[0108] Activation information for determining whether to infer at least one channel and / or trigger information for instructing to infer at least one channel may be included in the configuration information regarding at least one CSI-RS.

[0109] The location information of at least one CSI-RS may include location information for slots and resource blocks for which the processor (410) infers at least one channel in the time domain and / or the frequency domain according to one embodiment.

[0110] For example, the location information of at least one CSI-RS may include time domain location information for a first slot and frequency domain location information for a second resource block included in the first slot.

[0111] For example, the location information of at least one CSI-RS may include time domain location information for a second slot and frequency domain location information for a fifth resource block included in the second slot.

[0112] For example, the location information of at least one CSI-RS may include time domain location information for the fifth slot and frequency domain location information for the first resource block and the fourth resource block included in the fifth slot.

[0113] The location information of at least one CSI-RS may include location information for a starting resource block and an end resource block for a terminal to infer at least one channel in the frequency domain according to one embodiment.

[0114] The location information of at least one CSI-RS may include location information for a start resource block, an end resource block, and interval information of resource blocks for a terminal to infer at least one channel in the frequency domain according to one embodiment.

[0115] Here, the interval of the resource blocks can be set to a value smaller than the interval from the start resource block to the end resource block.

[0116] The position information of at least one CSI-RS may include position information for a start slot and an end slot for a terminal to infer at least one channel in the time domain according to one embodiment.

[0117] The position information of at least one CSI-RS may include position information for a start slot, an end slot, and interval information of slots for a terminal to infer at least one channel in the time domain according to one embodiment.

[0118] Here, the spacing between slots can be set to a value smaller than the spacing from the start slot to the end slot.

[0119] For example, the location information of at least one CSI-RS may include time domain location information for the first slot and frequency domain location information from the first resource block to the third resource block included in the first slot, but is not limited thereto.

[0120] The location information of at least one CSI-RS may include location information of a subcarrier and / or RE for the terminal to infer at least one channel in the time domain and / or the frequency domain according to one embodiment.

[0121] According to one embodiment, the location information of the CSI-RS may be transmitted from the base station to the terminal in the form of an index and / or a codebook.

[0122] The configuration information regarding at least one CSI-RS may include information indicating not to infer a channel in the time domain and / or frequency domain if the terminal does not need to infer a channel in the time domain and / or frequency domain.

[0123] According to one embodiment, the processor (410) can infer at least one channel through an artificial intelligence model based on the configuration information.

[0124] A processor (410) according to one embodiment may receive at least one CSI-RS from a base station to infer at least one channel through an artificial intelligence model.

[0125] According to one embodiment, the processor (410) can infer a channel through an artificial intelligence model based on at least one CSI-RS received from a base station or at least one of configuration information regarding the at least one CSI-RS.

[0126] According to one embodiment, the processor (410) can infer a channel through an artificial intelligence model based on at least one of at least one CSI-RS received from a base station, pattern information of at least one CSI-RS, or location information of at least one CSI-RS.

[0127] A processor (410) according to one embodiment may receive activation information from a base station for determining whether to infer at least one channel through at least one of MAC-CE or DCI, if the pattern information indicates that the period of at least one CSI-RS for inferring at least one channel is semi-persistent.

[0128] When the processor (410) according to one embodiment receives activation information from a base station to determine whether to infer at least one channel, the processor can infer at least one channel through an artificial intelligence model based on the activation information.

[0129] According to one embodiment, the processor (410) may infer at least one channel through an artificial intelligence model when receiving activation information from a base station that instructs inference of at least one channel.

[0130] In one embodiment, the processor (410) may not infer at least one channel through the artificial intelligence model when it receives activation information from the base station indicating not to infer a channel.

[0131] A processor (410) according to one embodiment can infer at least one channel from a time point when it receives activation information from a base station instructing it to infer at least one channel until a time point when it receives activation information from the base station instructing it not to infer at least one channel.

[0132] A processor (410) according to one embodiment may receive trigger information from a base station via DCI that instructs to infer at least one channel, if the pattern information indicates that the period of at least one CSI-RS for inferring at least one channel is aperiodic.

[0133] According to one embodiment, when a processor (410) receives trigger information from a base station, it can infer at least one channel through an artificial intelligence model based on the trigger information.

[0134] A processor (410) according to one embodiment can determine decay of an artificial intelligence model based on at least one inferred channel.

[0135] Degradation can mean that the performance of an AI model is degraded.

[0136] According to one embodiment, a processor (410) can receive at least one CSI-RS from a base station.

[0137] A processor (410) according to one embodiment can measure at least one channel based on at least one CSI-RS.

[0138] According to one embodiment, the processor (410) can determine the deterioration of the artificial intelligence model based on comparing at least one measured channel and at least one inferred channel.

[0139] A processor (410) according to one embodiment may compare at least one measured channel and at least one inferred channel based on at least one of squared generalized cosine similarity (SGCS), signal interference noise ratio (SINR), mean squared error (MSE), normalized mean squared error (NMSE), mean absolute error (MAE), or normalized mean absolute error (NMAE), but is not limited thereto.

[0140] SGCS is a method of measuring similarity by measuring the angle between two vectors to be compared.

[0141] SINR stands for signal to noise and interference ratio.

[0142] MSE stands for mean square error.

[0143] NMSE means MSE standardized to a value between 0 and 1.

[0144] MAE stands for Mean Absolute Error.

[0145] NMAE means MAE normalized to a value between 0 and 1.

[0146] A processor (410) according to one embodiment can obtain vector values ​​and / or matrix values ​​for at least one measured channel and at least one inferred channel.

[0147] Based on the output form of the artificial intelligence model according to one embodiment, a comparison method for at least one measured channel and at least one inferred channel can be determined.

[0148] According to one embodiment, the processor (410) can compare at least one measured channel and at least one inferred channel through at least one of SGCS, SINR, MSE, NMSE, NMAE, or MAE based on the acquired vector value and / or matrix value.

[0149] For example, if the processor (410) obtains vector values ​​for at least one measured channel and at least one inferred channel according to the output form of the artificial intelligence model, the processor (410) can compare the at least one measured channel and the at least one inferred channel based on the SGCS.

[0150] However, this is merely an example, and the comparison method corresponding to the output form of the artificial intelligence model may be determined based on the settings between the base station and the terminal. According to one embodiment, the processor (410) may determine the deterioration of the artificial intelligence model based on a preset threshold value.

[0151] According to one embodiment, the processor (410) can determine the deterioration of the artificial intelligence model based on comparing the comparison result value of at least one measured channel and at least one inferred channel with a threshold value.

[0152] According to one embodiment, the processor (410) can determine that deterioration has occurred in the artificial intelligence model when the comparison result value of at least one measured channel and at least one inferred channel is greater than or equal to a threshold value.

[0153] According to one embodiment, the processor (410) can determine that no deterioration has occurred in the artificial intelligence model if the comparison result value of at least one measured channel and at least one inferred channel is less than a threshold value.

[0154] The threshold value can be changed by at least one of the user, system settings, or the accuracy of the channel measurements required by the system.

[0155] A processor (410) according to one embodiment can update an artificial intelligence model based on degradation.

[0156] According to one embodiment, the processor (410) may update the artificial intelligence model, including but not limited to at least one of replacing the deteriorated artificial intelligence model with a new type of artificial intelligence model or re-training the deteriorated artificial intelligence model.

[0157] A new type of artificial intelligence model may mean an artificial intelligence model that has not deteriorated among at least one artificial intelligence model stored in a terminal according to one embodiment.

[0158] A new type of artificial intelligence model may mean at least one artificial intelligence model received by a terminal from a server according to one embodiment.

[0159] According to one embodiment, a processor (410) can input data previously received from a base station into an input terminal of an artificial intelligence model in which deterioration has occurred, thereby retraining the artificial intelligence model in which deterioration has occurred.

[0160] A processor (410) according to one embodiment can retrain an artificial intelligence model that has deteriorated based on data previously received from a base station and output corresponding to data previously received from the base station.

[0161] The data received from the base station may include, but is not limited to, information about at least one of the number of slots and / or the spacing between adjacent slots or the number of resource blocks and / or the spacing between adjacent resource blocks.

[0162] According to one embodiment, a processor (410) can transmit channel inference capability information for an updated artificial intelligence model to a base station via a transceiver (420).

[0163] Here, the channel inference capability information may include at least one of the following: an identifier of the updated AI model, the size of the updated AI model, the complexity of the updated AI model, information about the input of the updated AI model, or information about the output of the updated AI model, but is not limited thereto.

[0164] According to one embodiment, a processor (410) may receive configuration information regarding at least one CSI-RS used for channel inference of an updated artificial intelligence model from a base station.

[0165] According to one embodiment, the processor (410) can infer at least one channel through an updated artificial intelligence model based on configuration information regarding at least one CSI-RS received from a base station.

[0166] A processor (410) according to one embodiment can determine degradation of an updated model based on at least one inferred channel.

[0167] In FIG. 4, only essential components for explaining the operation of a terminal (400) according to one embodiment are illustrated, and the components included in the terminal (400) according to one embodiment are not limited as illustrated in FIG. 4.

[0168] FIG. 5 is a diagram for explaining an artificial intelligence model operated in a terminal according to one embodiment.

[0169] A terminal according to one embodiment may include an artificial intelligence (AI) model that performs calculations via a neural network.

[0170] A neural network (520) can be trained by receiving training data. Then, the trained neural network (520) can receive input data (510) as an input terminal (530), and the output terminal (550) can perform an operation to analyze the input data (510) and output output data (560), which is the intended result. The operation through the neural network can be performed through a hidden layer (540). In FIG. 5, for convenience, the hidden layer (540) is simplified and illustrated as being formed as a single layer, but the hidden layer (540) can be formed as a plurality of layers.

[0171] According to one embodiment, a processor can infer at least one channel through an artificial intelligence model based on configuration information by executing at least one instruction stored in a memory.

[0172] Although FIG. 5 is limited to describing an artificial intelligence model operated in a terminal according to one embodiment, a base station according to one embodiment may also include the artificial intelligence model illustrated in FIG. 5.

[0173] FIG. 6 is a diagram showing a terminal according to one embodiment that operates in conjunction with a server.

[0174] The server (600) may include a server, server system, server-based device, etc. that transmits and receives data to and from the terminal (400) through a communication unit and processes the data.

[0175] In the disclosed embodiment, the server (600) may include a communication unit (620) and a processor (610) that performs at least one command.

[0176] The server (600) may train an artificial intelligence model and store the trained artificial intelligence model.

[0177] In general, the terminal (400) may be limited in memory storage capacity, computational processing speed, and learning data set collection capability compared to the server (600). Therefore, operations requiring storage of large amounts of data and large amounts of computational power may be performed on the server (600), and then the necessary data and / or the artificial intelligence model being used may be transmitted to the terminal (400) via a communication network.

[0178] The terminal (400) can perform necessary operations quickly and easily by receiving and using necessary data and / or artificial intelligence models through a server without a processor having a large amount of memory and fast computing capabilities.

[0179] The server (600) may include a communication unit (620), a processor (610), and a database (DB) (630).

[0180] The communication unit (620) may include one or more components that enable communication with the terminal (400). The communication unit (620) includes at least one communication module, such as a short-range communication module, a wired communication module, a mobile communication module, a broadcast reception module, etc. Here, at least one communication module refers to a communication module that can transmit and receive data through a network that follows a communication standard, such as a tuner that performs broadcast reception, Bluetooth, WLAN (Wireless LAN) (Wi-Fi), Wibro (Wireless broadband), Wimax (World Interoperability for Microwave Access), CDMA, WCDMA, the Internet, 3G, 4G, 5G, and / or 6G millimeter wave (mmWAVE), etc.

[0181] Specifically, the mobile communication module included in the communication unit (620) can communicate with another device (e.g., a server (not shown)) located remotely via a communication network that complies with communication standards such as 3G, 4G, 5G, and / or 6G. Here, a communication module that communicates with a server (not shown) located remotely may be referred to as a 'remote communication module'.

[0182] The processor (610) controls the overall operation of the server (600). For example, the processor (610) may perform required operations by executing at least one instruction and at least one program of the server (600).

[0183] In addition, the DB (630) may include a memory (not shown), and may store at least one instruction, program, or data required for the server (600) to perform a predetermined operation within the memory (not shown). In addition, the DB (630) may store data required for the server (600) to perform operations according to a neural network.

[0184] Specifically, in the disclosed embodiment, the server (600) may store the neural network (520) described in FIG. 5. The neural network (520) may be stored in at least one of the processor (610) and the DB (630). The neural network (520) included in the server (600) may be a neural network that has completed training.

[0185] Additionally, the server (600) can transmit the neural network for which training has been completed to the transceiver (420) of the terminal (400) via the communication unit (620). Then, the terminal (400) can obtain and store the neural network for which training has been completed, and obtain the desired output data through the neural network.

[0186] Although FIG. 6 has been described limited to a terminal according to one embodiment that operates in conjunction with a server, a base station according to one embodiment may also operate in conjunction with a server as illustrated in FIG. 6.

[0187] FIG. 7 is a block diagram illustrating components of a base station according to one embodiment.

[0188] A base station (700) according to one embodiment may include a processor (710) and a transceiver (720).

[0189] A base station (700) according to one embodiment may further include a memory (not shown) that stores at least one command.

[0190] The memory may store various data, programs or applications for driving and controlling the base station (700) according to one embodiment. The memory may include, for example, a non-volatile memory including at least one of a flash memory type, a hard disk type, a multimedia card micro type, a card type memory (e.g., SD or XD memory, etc.), a ROM (Read-Only Memory), and an EEPROM (Electrically Erasable Programmable Read-Only Memory), a PROM (Programmable Read-Only Memory), and a volatile memory such as a RAM (Random Access Memory) or an SRAM (Static Random Access Memory).

[0191] The memory may store instructions, data structures, and program codes that can be read by the processor (710). In the following embodiments, the processor (710) may be implemented by executing instructions or codes of a program stored in the memory.

[0192] The processor (710) is a configuration that controls a series of processes so that the base station (700) operates according to the embodiments described below, and may be composed of one or more processors.

[0193] The processor (710) may be composed of hardware components that perform arithmetic, logic, and input / output operations and signal processing. One or more processors included in the processor (710) may be circuitry such as a System on Chip (SoC), an Integrated Circuit (IC), etc. The processor (710) may be composed of at least one of, for example, a Central Processing Unit (CPU), a microprocessor, a Graphic Processing Unit (GPU), Application Specific Integrated Circuits (ASICs), Digital Signal Processors (DSPs), Digital Signal Processing Devices (DSPDs), Programmable Logic Devices (PLDs), and Field Programmable Gate Arrays (FPGAs), but is not limited thereto.

[0194] The processor (710) can write data to memory, read data stored in memory, and process data according to predefined operation rules, particularly by executing a program or at least one instruction stored in memory.

[0195] The transceiver (720) can communicate with an external device or server through at least one wired or wireless communication network by means of the processor (710).

[0196] The transceiver (720) may include at least one short-range communication module that performs communication according to a communication standard such as Bluetooth, Wi-Fi, BLE (Bluetooth Low Energy), NFC / RFID, Wi-Fi Direct, UWB, or ZIGBEE, and a long-range communication module that performs communication with a server to support long-range communication according to a long-range communication standard. The long-range communication module may perform communication through a communication network according to a 3G, 4G, 5G, and / or 6G communication standard, or a network for Internet communication.

[0197] According to one embodiment, a processor (710) may receive channel inference capability information including information about an artificial intelligence model used for channel inference of the terminal from a terminal through a transceiver (720).

[0198] Channel inference capability information may include at least one of the following: an identifier of an AI model used for channel inference of the terminal, the size of the AI ​​model, the complexity of the AI ​​model, information regarding the input of the AI ​​model, or information regarding the output of the AI ​​model. However, this is not limited thereto.

[0199] According to one embodiment, a processor (710) may transmit, to a terminal, configuration information regarding at least one CSI-RS used for channel inference based on channel inference capability information through a transceiver (720).

[0200] According to one embodiment, the processor (710) may transmit configuration information regarding at least one CSI-RS to the terminal based on at least one of DCI, MAC-CE, or RRC signaling.

[0201] The configuration information regarding at least one CSI-RS may include, but is not limited to, at least one of pattern information of at least one CSI-RS for inferring at least one channel in the terminal or location information of at least one CSI-RS for inferring at least one channel.

[0202] According to one embodiment, the processor (710) may transmit configuration information regarding at least one CSI-RS to the terminal in the form of an index and / or codebook.

[0203] A processor (710) according to one embodiment may represent information regarding an output result of an artificial intelligence model identified based on channel inference capability information received from a terminal as a bit value or index.

[0204] For example, if the output result of the artificial intelligence model of the terminal is "predictable for the symbol of the second slot located four slots after the first slot," the processor (710) may indicate this as "index 1." However, the present invention is not limited thereto.

[0205] For example, if the output result of the artificial intelligence model of the terminal is "predictable symbols of the second slot located 6 slots after the first slot and symbols of the third slot located 4 slots after the second slot," the processor (710) may indicate this as "index 2." However, the present invention is not limited thereto.

[0206] According to one embodiment, a processor (710) may transmit an index indicating information about an output result of an artificial intelligence model as configuration information about at least one CSI-RS to a terminal.

[0207] For example, if the output result of the artificial intelligence model of the terminal is "predictable for the symbol of the second slot located four slots after the first slot," the processor (710) may indicate this as "index 1." The processor (710) may transmit "index 1" as configuration information regarding at least one CSI-RS to the terminal. The terminal that receives "index 1" may infer the symbol of the second slot located four slots after the first slot.

[0208] For example, if the output result of the artificial intelligence model of the terminal is "predictable symbols of the second slot located 6 slots after the first slot and symbols of the third slot located 4 slots after the second slot," the processor (710) may represent this as "index 2." The processor (710) may transmit "index 2" as configuration information regarding at least one CSI-RS to the terminal. The terminal that receives "index 2" may infer symbols of the second slot located 6 slots after the first slot and symbols of the third slot located 4 slots after the second slot.

[0209] The pattern information of at least one CSI-RS may include periodic information of at least one CSI-RS transmitted from a base station to a terminal according to one embodiment.

[0210] The periodic information of at least one CSI-RS may include at least one of periodic, semi-persistent, or aperiodic.

[0211] According to one embodiment, the processor (710) may transmit activation information for determining whether to infer at least one channel to the terminal through at least one of MAC-CE or DCI, if the pattern information indicates that the period of at least one CSI-RS for inferring at least one channel is semi-static.

[0212] According to one embodiment, the processor (710) may transmit trigger information to the terminal to infer at least one channel via DCI when the pattern information indicates that the period of at least one CSI-RS for inferring at least one channel is aperiodic.

[0213] The location information of at least one CSI-RS may include location information for slots and resource blocks for which the terminal infers at least one channel according to an embodiment in the time domain and / or the frequency domain.

[0214] For example, the location information of at least one CSI-RS may include time domain location information for a first slot and frequency domain location information for a second resource block included in the first slot.

[0215] For example, the location information of at least one CSI-RS may include time domain location information for a second slot and frequency domain location information for a fifth resource block included in the second slot.

[0216] For example, the location information of at least one CSI-RS may include time domain location information for the fifth slot and frequency domain location information for the first resource block and the fourth resource block included in the fifth slot.

[0217] The location information of at least one CSI-RS may include location information for a starting resource block and an end resource block for a terminal to infer at least one channel in the frequency domain according to one embodiment.

[0218] The location information of at least one CSI-RS may include location information for a start resource block, an end resource block, and interval information of resource blocks for a terminal to infer at least one channel in the frequency domain according to one embodiment.

[0219] Here, the interval of the resource blocks can be set to a value smaller than the interval from the start resource block to the end resource block.

[0220] The position information of at least one CSI-RS may include position information for a start slot and an end slot for a terminal to infer at least one channel in the time domain according to one embodiment.

[0221] The position information of at least one CSI-RS may include position information for a start slot, an end slot, and interval information of slots for a terminal to infer at least one channel in the time domain according to one embodiment.

[0222] Here, the spacing between slots can be set to a value smaller than the spacing from the start slot to the end slot.

[0223] For example, the location information of at least one CSI-RS may include time domain location information for the first slot and frequency domain location information from the first resource block to the third resource block included in the first slot, but is not limited thereto.

[0224] The location information of at least one CSI-RS may include location information of a subcarrier and / or RE for the terminal to infer at least one channel in the time domain and / or the frequency domain according to one embodiment.

[0225] According to one embodiment, the location information of the CSI-RS may be transmitted from the base station to the terminal in the form of an index and / or a codebook.

[0226] The configuration information regarding at least one CSI-RS may include information indicating not to infer a channel in the time domain and / or frequency domain if the terminal does not need to infer a channel in the time domain and / or frequency domain.

[0227] A processor (710) according to one embodiment may receive channel inference capability information according to an updated artificial intelligence model from a terminal.

[0228] Here, the channel inference capability information may include at least one of the following: an identifier of the updated AI model, the size of the updated AI model, the complexity of the updated AI model, information about the input of the updated AI model, or information about the output of the updated AI model, but is not limited thereto.

[0229] According to one embodiment, a processor (710) may transmit configuration information regarding at least one CSI-RS used for channel inference of an updated artificial intelligence model to a terminal.

[0230] According to one embodiment, a processor (710) may transmit configuration information regarding at least one CSI-RS used for channel inference of an updated artificial intelligence model to a terminal based on RRC signaling.

[0231] For example, the processor (710) may transmit configuration information regarding at least one CSI-RS used for channel inference of an updated artificial intelligence model based on RRC reconfiguration to the terminal.

[0232] According to one embodiment, configuration information regarding at least one CSI-RS may be provided for each terminal. However, this is only an embodiment, and the configuration information regarding at least one CSI-RS may be cell-specific information.

[0233] For example, the processor (710) may transmit configuration information regarding at least one CSI-RS used for channel inference of an updated artificial intelligence model to at least one terminal connected to a cell. A first terminal and a second terminal connected to a first cell may receive configuration information regarding at least one CSI-RS used for channel inference of an updated artificial intelligence model.

[0234] In FIG. 7, only essential components for explaining the operation of a base station (700) according to one embodiment are illustrated, and the components included in the base station (700) according to one embodiment are not limited to those illustrated in FIG. 7.

[0235] Figure 8 is a flowchart illustrating a method of operating a terminal according to one embodiment.

[0236] In step 810, the terminal may transmit channel inference information including information about an artificial intelligence model used for channel inference of the terminal to the base station.

[0237] A terminal according to one embodiment can store at least one artificial intelligence model in memory.

[0238] A detailed description of the artificial intelligence model operated in the terminal according to one embodiment is described above in FIG. 5, so redundant description is omitted.

[0239] According to one embodiment, a terminal can receive at least one artificial intelligence model for which learning has been completed from a server.

[0240] A specific description of how a terminal according to one embodiment receives at least one artificial intelligence model for which learning has been completed from a server has been described above in FIG. 6, so a redundant description is omitted.

[0241] A terminal according to one embodiment can store at least one artificial intelligence model, for which learning has been completed and received from a server, in memory.

[0242] A terminal according to one embodiment can control an operation performed in the terminal according to one embodiment through an artificial intelligence model.

[0243] Channel inference capability information may include at least one of the following: an identifier of the AI ​​model, the size of the AI ​​model, the complexity of the AI ​​model, information about the input of the AI ​​model, or information about the output of the AI ​​model, but is not limited thereto.

[0244] According to one embodiment, a terminal may assign an identifier to each of at least one artificial intelligence models.

[0245] A terminal according to one embodiment may assign an identifier to each of at least one artificial intelligence model based on at least one of indexing or bitmap.

[0246] According to one embodiment, a terminal may assign an identifier to each of at least one artificial intelligence model according to the purpose of use of the at least one artificial intelligence model.

[0247] For example, if the first artificial intelligence model stored in the terminal is an artificial intelligence model for beam management, the terminal may assign “identifier 1” to the first artificial intelligence model.

[0248] For example, if the second artificial intelligence model is an artificial intelligence model for positioning accuracy that accurately measures the location of the terminal, the terminal may assign "identifier 2" to the second artificial intelligence model.

[0249] For example, the terminal may assign “identifier 3” to the third artificial intelligence model if the third artificial intelligence model is an artificial intelligence model for channel inference.

[0250] According to one embodiment, a terminal can distinguish at least one artificial intelligence model based on at least one of the number of layers of the artificial intelligence model or learning data.

[0251] According to one embodiment, a terminal may assign an identifier to each of at least one artificial intelligence model included in at least one artificial intelligence model assigned an identifier according to the purpose of use.

[0252] For example, the terminal can distinguish at least one artificial intelligence model included in the third artificial intelligence model assigned “identifier 3”.

[0253] According to one embodiment, the terminal may assign an identifier to each of at least one artificial intelligence model included in the third artificial intelligence model.

[0254] According to one embodiment, a terminal may assign an identifier of “identifier 3, type A” to an artificial intelligence model A included in a third artificial intelligence model.

[0255] According to one embodiment, a terminal may assign an identifier of “identifier 3, type B” to an artificial intelligence model B included in a third artificial intelligence model.

[0256] However, this is merely an example, and the method by which a terminal assigns an identifier to at least one AI model according to one embodiment is not limited to this. In another example, the identifier indicating that the third AI model is an AI model for channel inference and the identifier indicating the type of AI model used for channel inference may exist as separate parameters or fields.

[0257] However, this is only an example of a terminal according to one embodiment assigning an identifier to at least one artificial intelligence model, and is not limited thereto.

[0258] The size of the AI ​​model may include at least one parameter value, including the number of layers used for learning the AI ​​model or weights, but is not limited thereto.

[0259] The complexity of an AI model can refer to the amount of computation required to train the AI ​​model.

[0260] Information about the input of an artificial intelligence model can refer to data entered as input into the input stage of an artificial intelligence model.

[0261] For example, information about the input of an artificial intelligence model may include, but is not limited to, vector values ​​and / or matrix values ​​for at least one of the number of slots in the time domain or the interval between adjacent slots.

[0262] For example, information about the input of the artificial intelligence model may include, but is not limited to, vector values ​​and / or matrix values ​​for at least one of the number of resource blocks in the frequency domain or the spacing between adjacent resource blocks.

[0263] For example, information about the input of the artificial intelligence model may include, but is not limited to, vector values ​​and / or matrix values ​​for at least one of the number of sub-carriers in the frequency domain or the spacing between adjacent sub-carriers.

[0264] Information about the output results of an artificial intelligence model can mean the output according to the input of the artificial intelligence model.

[0265] For example, information regarding the output result of an artificial intelligence model may include information indicating that "the symbol of the second slot located four slots after the first slot is predictable", which is an output corresponding to the input, when the number of slots and / or the spacing between adjacent slots is input to the input terminal of the artificial intelligence model. However, this is not limited thereto.

[0266] For example, information about the output result of the artificial intelligence model may include information indicating that "the 6th symbol included in the 2nd slot is predictable", which is an output corresponding to the input, when the number of slots and / or the spacing between adjacent slots is input to the input terminal of the artificial intelligence model.

[0267] For example, information about the output result of an artificial intelligence model may include information indicating that "the second symbol to the eighth symbol included in the first slot can be predicted", which is an output corresponding to the input, when the number of slots and / or the spacing between adjacent slots is input to the input terminal of the artificial intelligence model.

[0268] For example, information about the output result of the artificial intelligence model may include information indicating that, when the number of slots and / or the spacing between adjacent slots is input to the input terminal of the artificial intelligence model, the output corresponding to the input is "predictable from the second symbol to the fourth symbol included in the second slot and from the sixth symbol to the eighth symbol included in the second slot."

[0269] For example, information about the output result of the artificial intelligence model may include information indicating that, when the number of slots and / or the spacing between adjacent slots is input to the input terminal of the artificial intelligence model, "at least one symbol included in the third slot can be predicted in units of two symbols starting from the first symbol included in the third slot," which is an output corresponding to the input.

[0270] For example, information regarding the output result of an artificial intelligence model may include information indicating that, when the number of slots and / or the spacing between adjacent slots is input to the input terminal of the artificial intelligence model, "the symbol of the second slot located 6 slots after the first slot and the symbol of the third slot located 4 slots after the second slot can be predicted," which is an output corresponding to the input. However, this is not limited thereto.

[0271] For example, information regarding the output result of an artificial intelligence model may include information indicating that "symbols can be predicted in units of two resource blocks from the first resource block", which is an output corresponding to the input, when the number of resource blocks and / or the spacing between adjacent resource blocks is input to the input stage of the artificial intelligence model. However, this is not limited thereto.

[0272] For example, information about the output result of the artificial intelligence model may include information indicating that, when at least one of the number of slots and / or the spacing between adjacent slots or the number of resource blocks and / or the spacing between adjacent resource blocks is input as an input to the input terminal of the artificial intelligence model, "in the second slot located four slots after the first slot, a symbol can be predicted in units of two resource blocks from the first resource block," which is an output corresponding to the input. However, the present invention is not limited thereto.

[0273] According to one embodiment, a terminal may transmit channel inference capability information to a base station based on at least one of uplink control information (UCI), medium access control-control element (MAC-CE), or radio resource control signaling (RRC signaling).

[0274] In step 820, the terminal may receive configuration information about at least one CSI-RS used for channel inference from the base station based on channel inference capability information.

[0275] According to one embodiment, a terminal may receive configuration information regarding at least one CSI-RS from a base station based on at least one of downlink control information (DCI), MAC-CE, or RRC signaling.

[0276] The configuration information regarding at least one CSI-RS may include, but is not limited to, at least one of pattern information of at least one CSI-RS for inferring at least one channel or location information of at least one CSI-RS for inferring at least one channel.

[0277] The configuration information regarding at least one CSI-RS may be transmitted from the base station to another terminal in one embodiment in the form of an index and / or a codebook.

[0278] The pattern information of at least one CSI-RS may include periodic information of at least one CSI-RS transmitted from a base station to a terminal.

[0279] The periodic information of at least one CSI-RS may include at least one of periodic, semi-persistent, or aperiodic.

[0280] According to one embodiment, a terminal may receive activation information from a base station for determining whether to infer at least one channel through at least one of MAC-CE or DCI, when the pattern information indicates that the period of at least one CSI-RS for inferring at least one channel is semi-static.

[0281] According to one embodiment, a terminal may receive trigger information from a base station via DCI instructing to infer at least one channel, if the pattern information indicates that the period of at least one CSI-RS for inferring at least one channel is aperiodic.

[0282] Activation information for determining whether to infer at least one channel and / or trigger information for instructing to infer at least one channel may be included in the configuration information regarding at least one CSI-RS.

[0283] The location information of at least one CSI-RS may include location information for slots and resource blocks for which the terminal infers at least one channel according to an embodiment in the time domain and / or the frequency domain.

[0284] For example, the location information of at least one CSI-RS may include time domain location information for a first slot and frequency domain location information for a second resource block included in the first slot.

[0285] For example, the location information of at least one CSI-RS may include time domain location information for a second slot and frequency domain location information for a fifth resource block included in the second slot.

[0286] For example, the location information of at least one CSI-RS may include time domain location information for the fifth slot and frequency domain location information for the first resource block and the fourth resource block included in the fifth slot.

[0287] The location information of at least one CSI-RS may include location information for a starting resource block and an end resource block for a terminal to infer at least one channel in the frequency domain according to one embodiment.

[0288] The location information of at least one CSI-RS may include location information for a start resource block, an end resource block, and interval information of resource blocks for a terminal to infer at least one channel in the frequency domain according to one embodiment.

[0289] Here, the interval of the resource blocks can be set to a value smaller than the interval from the start resource block to the end resource block.

[0290] The position information of at least one CSI-RS may include position information for a start slot and an end slot for a terminal to infer at least one channel in the time domain according to one embodiment.

[0291] The position information of at least one CSI-RS may include position information for a start slot, an end slot, and interval information of slots for a terminal to infer at least one channel in the time domain according to one embodiment.

[0292] Here, the spacing between slots can be set to a value smaller than the spacing from the start slot to the end slot.

[0293] For example, the location information of at least one CSI-RS may include time domain location information for the first slot and frequency domain location information from the first resource block to the third resource block included in the first slot, but is not limited thereto.

[0294] The location information of at least one CSI-RS may include location information of a subcarrier and / or RE for the terminal to infer at least one channel in the time domain and / or the frequency domain according to one embodiment.

[0295] According to one embodiment, the location information of the CSI-RS may be transmitted from the base station to the terminal in the form of an index and / or a codebook.

[0296] The configuration information regarding at least one CSI-RS may include information indicating not to infer a channel in the time domain and / or frequency domain if the terminal does not need to infer a channel in the time domain and / or frequency domain.

[0297] At step 830, the terminal can infer at least one channel through an artificial intelligence model based on the configuration information.

[0298] According to one embodiment, a terminal may receive at least one CSI-RS from a base station for inferring at least one channel through an artificial intelligence model.

[0299] According to one embodiment, a terminal may infer a channel through an artificial intelligence model based on at least one CSI-RS received from a base station or at least one of configuration information regarding at least one CSI-RS.

[0300] According to one embodiment, a terminal can infer a channel through an artificial intelligence model based on at least one of at least one CSI-RS received from a base station, pattern information of at least one CSI-RS, or location information of at least one CSI-RS.

[0301] According to one embodiment, a terminal may receive activation information from a base station, through at least one of MAC-CE or DCI, for determining whether to infer at least one channel, when the pattern information indicates that the period of at least one CSI-RS for inferring at least one channel is semi-persistent.

[0302] According to one embodiment, when a terminal receives activation information from a base station to determine whether to infer at least one channel, the terminal can infer at least one channel through an artificial intelligence model based on the activation information.

[0303] According to one embodiment, a terminal may infer at least one channel through an artificial intelligence model when receiving activation information from a base station that instructs inference of at least one channel.

[0304] In one embodiment, a terminal may not infer at least one channel through an artificial intelligence model when it receives activation information from a base station indicating not to infer a channel.

[0305] According to one embodiment, a terminal can infer at least one channel from a time point when it receives activation information from a base station instructing it to infer at least one channel until a time point when it receives activation information from the base station instructing it not to infer at least one channel.

[0306] According to one embodiment, a terminal may receive trigger information from a base station via DCI instructing to infer at least one channel, if the pattern information indicates that the period of at least one CSI-RS for inferring at least one channel is aperiodic.

[0307] According to one embodiment, when a terminal receives trigger information from a base station, the terminal can infer at least one channel through an artificial intelligence model based on the trigger information.

[0308] At step 840, the terminal can determine decay of the artificial intelligence model based on at least one inferred channel.

[0309] Degradation can mean that the performance of an AI model is degraded.

[0310] A specific method for determining deterioration of an artificial intelligence model based on at least one inferred channel by a terminal according to one embodiment will be described with reference to FIG. 9.

[0311] Figure 9 is a flowchart illustrating a method of operating a terminal according to one embodiment.

[0312] Steps 830 and 840 of FIG. 9 correspond to steps 830 and 840 of FIG. 8, respectively.

[0313] In step 842, the terminal may receive at least one CSI-RS from the base station.

[0314] In step 844, the terminal may measure at least one channel based on at least one CSI-RS.

[0315] In step 846, the terminal can determine degradation of the artificial intelligence model based on comparing at least one measured channel and at least one inferred channel.

[0316] According to one embodiment, a terminal may compare at least one measured channel and at least one inferred channel based on at least one of squared generalized cosine similarity (SGCS), signal interference noise ratio (SINR), mean squared error (MSE), normalized mean squared error (NMSE), mean absolute error (MAE), or normalized mean absolute error (NMAE), but is not limited thereto.

[0317] SGCS is a method of measuring similarity by measuring the angle between two vectors to be compared.

[0318] SINR stands for signal to noise and interference ratio.

[0319] MSE stands for mean square error.

[0320] NMSE means MSE standardized to a value between 0 and 1.

[0321] MAE stands for Mean Absolute Error.

[0322] NMAE means MAE normalized to a value between 0 and 1.

[0323] A terminal according to one embodiment can obtain vector values ​​and / or matrix values ​​for at least one measured channel and at least one inferred channel.

[0324] Based on the output form of the artificial intelligence model according to one embodiment, a comparison method for at least one measured channel and at least one inferred channel can be determined.

[0325] According to one embodiment, the terminal may compare at least one measured channel and at least one inferred channel through at least one of SGCS, SINR, MSE, NMSE, NMAE, or MAE based on the acquired vector value and / or matrix value.

[0326] For example, if the terminal obtains vector values ​​for at least one measured channel and at least one inferred channel according to the output form of the artificial intelligence model, the terminal can compare the at least one measured channel and the at least one inferred channel based on the SGCS.

[0327] However, this is merely an example, and the comparison method corresponding to the output form of the AI ​​model may be determined based on the settings between the base station and the terminal. In one embodiment, the terminal may determine AI model deterioration based on a preset threshold.

[0328] According to one embodiment, a terminal can determine deterioration of an artificial intelligence model based on comparing a comparison result value of at least one measured channel and at least one inferred channel with a threshold value.

[0329] According to one embodiment, a terminal may determine that a degradation has occurred in an artificial intelligence model when a comparison result value of at least one measured channel and at least one inferred channel is greater than or equal to a threshold value.

[0330] According to one embodiment, a terminal may determine that no degradation has occurred in an artificial intelligence model if a comparison result value of at least one measured channel and at least one inferred channel is less than a threshold value.

[0331] The threshold value can be changed by at least one of the user, system settings, or the accuracy of the channel measurements required by the system.

[0332] Fig. 10 is a flowchart illustrating an operation method of a terminal according to one embodiment.

[0333] Step 840 of FIG. 10 corresponds to step 840 of FIG. 8.

[0334] At step 850, the terminal can update the artificial intelligence model based on the degradation.

[0335] Updating an AI model may include, but is not limited to, at least one of replacing a degraded AI model with a new type of AI model or re-training the degraded AI model.

[0336] A new type of artificial intelligence model may mean an artificial intelligence model that has not deteriorated among at least one artificial intelligence model stored in a terminal according to one embodiment.

[0337] A new type of artificial intelligence model may mean at least one artificial intelligence model received by a terminal from a server according to one embodiment.

[0338] According to one embodiment, a terminal can retrain an artificial intelligence model in which deterioration has occurred by inputting data previously received from a base station into an input terminal of an artificial intelligence model in which deterioration has occurred.

[0339] In one embodiment, a terminal can retrain an artificial intelligence model that has experienced degradation based on data previously received from a base station and output corresponding to data previously received from the base station.

[0340] The data received from the base station may include, but is not limited to, information about at least one of the number of slots and / or the spacing between adjacent slots or the number of resource blocks and / or the spacing between adjacent resource blocks.

[0341] At step 860, the terminal may transmit channel inference capability information for the updated artificial intelligence model to the base station.

[0342] Here, the channel inference capability information may include at least one of the following: an identifier of the updated AI model, the size of the updated AI model, the complexity of the updated AI model, information about the input of the updated AI model, or information about the output of the updated AI model, but is not limited thereto.

[0343] In step 870, the terminal may receive configuration information regarding at least one CSI-RS used for channel inference of an updated artificial intelligence model from the base station.

[0344] According to one embodiment, a terminal can infer at least one channel through an updated artificial intelligence model based on configuration information regarding at least one CSI-RS received from a base station.

[0345] A terminal according to one embodiment can determine degradation of an updated model based on at least one inferred channel.

[0346] Fig. 11 is a flowchart illustrating an operation method of a base station according to one embodiment.

[0347] In step 1110, the base station can receive channel inference capability information from the terminal, which includes information about an artificial intelligence model used for channel inference of the terminal.

[0348] Channel inference capability information may include at least one of the following: an identifier of an AI model used for channel inference of the terminal, the size of the AI ​​model, the complexity of the AI ​​model, information regarding the input of the AI ​​model, or information regarding the output of the AI ​​model. However, this is not limited thereto.

[0349] In step 1120, the base station may transmit configuration information regarding at least one CSI-RS used for channel inference to the terminal based on channel inference capability information.

[0350] According to one embodiment, a base station may transmit configuration information regarding at least one CSI-RS to a terminal based on at least one of DCI, MAC-CE, or RRC signaling.

[0351] The configuration information regarding at least one CSI-RS may include, but is not limited to, at least one of pattern information of at least one CSI-RS for inferring at least one channel in the terminal or location information of at least one CSI-RS for inferring at least one channel.

[0352] According to one embodiment, a base station may transmit configuration information regarding at least one CSI-RS to a terminal in the form of an index and / or a codebook.

[0353] According to one embodiment, a base station may represent information regarding an output result of an artificial intelligence model identified based on channel inference capability information received from a terminal as a bit value or index.

[0354] For example, if the output result of the terminal's artificial intelligence model is "predictable for the symbol of the second slot located four slots after the first slot," the base station may indicate this as "index 1." However, this is not limited thereto.

[0355] For example, if the output result of the terminal's artificial intelligence model is "predictable for the symbol of the second slot located 6 slots after the first slot and the symbol of the third slot located 4 slots after the second slot," the base station may indicate this as "index 2." However, this is not limited thereto.

[0356] For example, if the output result of the terminal's artificial intelligence model is "predictable for the 6th symbol included in the 2nd slot," the base station may indicate this as "index 3." However, this is not limited thereto.

[0357] For example, if the output result of the artificial intelligence model of the terminal is "the second symbol to the eighth symbol included in the first slot can be predicted," the base station may indicate this as "index 4." However, this is not limited thereto. According to one embodiment, the base station may transmit to the terminal an index indicating information about the output result of the artificial intelligence model as configuration information about at least one CSI-RS.

[0358] For example, if the output result of the artificial intelligence model of the terminal is "predictable for the symbol of the second slot located four slots after the first slot," the base station can indicate this as "index 1." The base station can transmit "index 1" as configuration information regarding at least one CSI-RS to the terminal. The terminal that receives "index 1" can infer the symbol of the second slot located four slots after the first slot.

[0359] For example, if the output result of the artificial intelligence model of the terminal is "predictable symbols of the second slot located 6 slots after the first slot and symbols of the third slot located 4 slots after the second slot," the base station may indicate this as "index 2." The base station may transmit "index 2" as configuration information regarding at least one CSI-RS to the terminal. The terminal that receives "index 2" can "infer symbols of the second slot located 6 slots after the first slot and symbols of the third slot located 4 slots after the second slot."

[0360] For example, if the output result of the terminal's artificial intelligence model is "predictable for the 6th symbol included in the 2nd slot," the base station can indicate this as "index 3." The base station can transmit "index 3" to the terminal as configuration information regarding at least one CSI-RS. The terminal that receives "index 3" can infer the 6th symbol included in the 2nd slot.

[0361] For example, if the output result of the artificial intelligence model of the terminal is "the second to eighth symbols included in the first slot can be predicted," the base station may indicate this as "index 4." The base station may transmit "index 4" as configuration information regarding at least one CSI-RS to the terminal. The terminal that receives "index 4" may infer the second to eighth symbols included in the first slot. However, this is not limited thereto.

[0362] The pattern information of at least one CSI-RS may include periodic information of at least one CSI-RS transmitted from a base station to a terminal according to one embodiment. The periodic information of at least one CSI-RS may include at least one of periodic, semi-persistent, or aperiodic.

[0363] In one embodiment, a base station may transmit activation information for determining whether to infer at least one channel to a terminal via at least one of MAC-CE or DCI, when the pattern information indicates that the period of at least one CSI-RS for inferring at least one channel is semi-static.

[0364] According to one embodiment, a base station may transmit trigger information to a terminal through DCI, which instructs the terminal to infer at least one channel, if the pattern information indicates that the period of at least one CSI-RS for inferring at least one channel is aperiodic.

[0365] The location information of at least one CSI-RS may include location information for slots and resource blocks for which the terminal infers at least one channel according to an embodiment in the time domain and / or the frequency domain.

[0366] For example, the location information of at least one CSI-RS may include time domain location information for a first slot and frequency domain location information for a second resource block included in the first slot.

[0367] For example, the location information of at least one CSI-RS may include time domain location information for a second slot and frequency domain location information for a fifth resource block included in the second slot.

[0368] For example, the location information of at least one CSI-RS may include time domain location information for the fifth slot and frequency domain location information for the first resource block and the fourth resource block included in the fifth slot.

[0369] The location information of at least one CSI-RS may include location information for a starting resource block and an end resource block for a terminal to infer at least one channel in the frequency domain according to one embodiment.

[0370] The location information of at least one CSI-RS may include location information for a start resource block, an end resource block, and interval information of resource blocks for a terminal to infer at least one channel in the frequency domain according to one embodiment.

[0371] Here, the interval of the resource blocks can be set to a value smaller than the interval from the start resource block to the end resource block.

[0372] The position information of at least one CSI-RS may include position information for a start slot and an end slot for a terminal to infer at least one channel in the time domain according to one embodiment.

[0373] The position information of at least one CSI-RS may include position information for a start slot, an end slot, and interval information of slots for a terminal to infer at least one channel in the time domain according to one embodiment.

[0374] Here, the spacing between slots can be set to a value smaller than the spacing from the start slot to the end slot.

[0375] For example, the location information of at least one CSI-RS may include time domain location information for the first slot and frequency domain location information from the first resource block to the third resource block included in the first slot, but is not limited thereto.

[0376] The location information of at least one CSI-RS may include location information of a subcarrier and / or RE for the terminal to infer at least one channel in the time domain and / or the frequency domain according to one embodiment.

[0377] According to one embodiment, the location information of the CSI-RS may be transmitted from the base station to the terminal in the form of an index and / or a codebook.

[0378] The configuration information regarding at least one CSI-RS may include information indicating not to infer a channel in the time domain and / or frequency domain if the terminal does not need to infer a channel in the time domain and / or frequency domain.

[0379] Fig. 12 is a flowchart illustrating an operation method of a base station according to one embodiment.

[0380] Step 1120 of FIG. 12 corresponds to step 1120 of FIG. 11.

[0381] At step 1130, the base station can receive channel inference capability information according to the updated artificial intelligence model from the terminal.

[0382] Here, the channel inference capability information may include at least one of the following: an identifier of the updated AI model, the size of the updated AI model, the complexity of the updated AI model, information about the input of the updated AI model, or information about the output of the updated AI model, but is not limited thereto.

[0383] In step 1140, the base station may transmit configuration information regarding at least one CSI-RS used for channel inference of the updated artificial intelligence model to the terminal.

[0384] According to one embodiment, a base station may transmit configuration information regarding at least one CSI-RS used for channel inference of an updated artificial intelligence model to a terminal based on RRC signaling.

[0385] For example, the base station may transmit configuration information regarding at least one CSI-RS used for channel inference of an updated artificial intelligence model based on RRC reconfiguration to the terminal.

[0386] According to one embodiment, configuration information regarding at least one CSI-RS may be provided for each terminal. However, this is only an embodiment, and the configuration information regarding at least one CSI-RS may be cell-specific information.

[0387] For example, a base station may transmit configuration information regarding at least one CSI-RS used for channel inference of an updated artificial intelligence model to at least one terminal connected to the cell. A first terminal and a second terminal connected to the first cell may receive configuration information regarding at least one CSI-RS used for channel inference of the updated artificial intelligence model.

[0388] FIG. 13 is a flowchart illustrating an operation method of a terminal according to one embodiment and a base station according to one embodiment.

[0389] At step 1310, the terminal may transmit channel inference capability information to the base station.

[0390] Channel inference capability information may include at least one of the following: an identifier of the AI ​​model, the size of the AI ​​model, the complexity of the AI ​​model, information about the input of the AI ​​model, or information about the output of the AI ​​model, but is not limited thereto.

[0391] A terminal according to one embodiment may transmit channel inference capability information to a base station based on at least one of UCI, MAC-CE, or RRC signaling.

[0392] At step 1320, the base station may transmit channel inference instruction information to the terminal.

[0393] The channel inference instruction information may include configuration information regarding at least one CSI-RS used for channel inference in the terminal.

[0394] According to one embodiment, a base station may transmit configuration information regarding at least one CSI-RS to a terminal based on at least one of DCI, MAC-CE, or RRC signaling.

[0395] The channel inference instruction information may include, but is not limited to, at least one of pattern information of at least one CSI-RS for inferring at least one channel or position information of at least one CSI-RS for inferring at least one channel.

[0396] In step 1330, the terminal can infer at least one channel through an artificial intelligence model based on channel inference instruction information.

[0397] According to one embodiment, a terminal can infer a channel through an artificial intelligence model based on at least one CSI-RS or channel inference indication information received from a base station.

[0398] In step 1340, the terminal may transmit information about at least one inferred channel to the base station.

[0399] At step 1350, the base station can determine degradation of the artificial intelligence model based on information about at least one received channel.

[0400] According to one embodiment, a base station may determine degradation of an artificial intelligence model through at least one of SGCS, SINR, MSE, NMSE, MAE, or NMAE based on information about at least one channel received from a terminal, but is not limited thereto.

[0401] At step 1360, the base station may transmit update instruction information for the artificial intelligence model to the terminal based on the deterioration.

[0402] In one embodiment, a base station may determine whether an update of an artificial intelligence model is necessary based on the determined deterioration of the artificial intelligence model.

[0403] In one embodiment, a base station may determine whether an update of an artificial intelligence model is necessary based on the determined deterioration of the artificial intelligence model and a preset threshold value.

[0404] The threshold value can be changed by at least one of the user, system settings, or the accuracy of the channel measurements required by the system.

[0405] In one embodiment, a base station may transmit update instruction information for an artificial intelligence model to a terminal when an update of the artificial intelligence model is required.

[0406] The update instruction information for the artificial intelligence model may include, but is not limited to, at least one of information instructing to replace the artificial intelligence model in which deterioration has occurred with a new type of artificial intelligence model or information instructing to retrain the artificial intelligence model in which deterioration has occurred.

[0407] The operation of the terminal described in Fig. 13 can be performed by the processor and transceiver included in the terminal described above in Fig. 4.

[0408] The operation of the base station described in FIG. 13 can be performed by the processor and transceiver included in the base station described above in FIG. 7.

[0409] FIG. 14 is a flowchart illustrating an operation method of a terminal according to one embodiment and a base station according to one embodiment.

[0410] At step 1410, the base station may transmit channel inference capability information to the terminal.

[0411] A base station according to one embodiment may store at least one artificial intelligence model in memory.

[0412] A detailed description of the artificial intelligence model used in the base station according to one embodiment is described above in FIG. 5, so a redundant description is omitted.

[0413] According to one embodiment, a base station can receive at least one artificial intelligence model for which learning has been completed from a server.

[0414] A specific description of how a base station according to one embodiment receives at least one artificial intelligence model for which learning has been completed from a server has been described above in FIG. 6, so a redundant description is omitted.

[0415] According to one embodiment, a base station can store in memory at least one artificial intelligence model for which learning has been completed and received from a server.

[0416] A base station according to one embodiment can control operations performed in a terminal according to one embodiment through an artificial intelligence model.

[0417] Channel inference capability information may include at least one of the following: an identifier of the AI ​​model, the size of the AI ​​model, the complexity of the AI ​​model, information about the input of the AI ​​model, or information about the output of the AI ​​model, but is not limited thereto.

[0418] According to one embodiment, a base station may assign an identifier to each of at least one artificial intelligence model.

[0419] In one embodiment, the base station may assign an identifier to each of at least one artificial intelligence model based on at least one of indexing or bitmap.

[0420] According to one embodiment, a base station may assign an identifier to each of at least one artificial intelligence model according to the purpose of use of the at least one artificial intelligence model.

[0421] The size of the AI ​​model may include at least one parameter value, including the number of layers used for learning the AI ​​model or weights, but is not limited thereto.

[0422] The complexity of an AI model can refer to the amount of computation required to train the AI ​​model, but is not limited to this.

[0423] Information about the input of an artificial intelligence model can refer to data entered as input into the input stage when training an artificial intelligence model.

[0424] Information about the output results of an artificial intelligence model can mean the output according to the input of the artificial intelligence model.

[0425] A base station according to one embodiment may transmit channel inference capability information to a terminal based on at least one of DCI, MAC-CE, or RRC signaling.

[0426] At step 1420, the terminal may transmit channel inference instruction information to the base station.

[0427] The channel inference instruction information may include configuration information regarding at least one sounding-reference signal (SRS) used for channel inference at the base station.

[0428] Channel inference instruction information may include information about a channel measured by a terminal based on a CSI-RS received from a base station according to one embodiment.

[0429] According to one embodiment, a terminal may transmit channel inference indication information to a base station based on at least one of UCI, MAC-CE, or RRC signaling.

[0430] According to one embodiment, a terminal may transmit to a base station configuration information about at least one SRS used for channel inference at the base station based on at least one of UCI, MAC-CE or RRC signaling and / or information about a channel measured based on CSI-RS.

[0431] The configuration information regarding at least one SRS may include, but is not limited to, at least one of pattern information of at least one SRS for inferring at least one channel or location information of at least one SRS for inferring at least one channel.

[0432] The pattern information of at least one SRS may include periodic information of at least one SRS transmitted from the terminal to the base station.

[0433] The periodic information of at least one SRS may include at least one of periodic, semi-persistent, or aperiodic.

[0434] According to one embodiment, a terminal may transmit activation information to a base station via at least one of MAC-CE or UCI to determine whether to infer at least one channel, if the pattern information indicates that the period of at least one SRS for inferring at least one channel is semi-static.

[0435] According to one embodiment, a terminal may transmit trigger information to a base station via UCI to infer at least one channel, if the pattern information indicates that the period of at least one SRS for inferring at least one channel is aperiodic.

[0436] Activation information for determining whether to infer at least one channel and / or trigger information for instructing to infer at least one channel may be included in the configuration information regarding at least one SRS.

[0437] The location information of at least one SRS may include location information for slots and resource blocks for which the base station infers at least one channel in the time domain and / or the frequency domain, according to one embodiment.

[0438] The location information of at least one SRS may include location information of a sub-carrier and / or RE for the terminal to infer at least one channel according to an embodiment in the time domain and / or the frequency domain.

[0439] According to one embodiment, the location information of the SRS may be transmitted from the terminal to the base station in the form of an index and / or a codebook.

[0440] The configuration information regarding at least one SRS may include information indicating that the base station should not infer the channel in the time domain and / or the frequency domain if there is no need to infer the channel in the time domain and / or the frequency domain.

[0441] At step 1430, the base station can infer at least one channel through an artificial intelligence model.

[0442] According to one embodiment, a base station can infer at least one channel through an artificial intelligence model based on channel inference instruction information received from a terminal.

[0443] According to one embodiment, a base station can infer at least one channel through an artificial intelligence model based on configuration information regarding at least one SRS received from a terminal and / or information regarding a channel measured by the terminal based on CSI-RS.

[0444] According to one embodiment, a base station can infer a channel through an artificial intelligence model based on at least one of at least one SRS received from a terminal, pattern information of at least one SRS, or location information of at least one SRS.

[0445] In one embodiment, a base station may receive activation information from a terminal for determining whether to infer at least one channel through at least one of MAC-CE or UCI, if the pattern information indicates that the period of at least one SRS for inferring at least one channel is semi-persistent.

[0446] In one embodiment, when a base station receives activation information from a terminal to determine whether to infer at least one channel, the base station can infer at least one channel through an artificial intelligence model based on the activation information.

[0447] In one embodiment, a base station may infer at least one channel through an artificial intelligence model when receiving activation information from a terminal that instructs inference of at least one channel.

[0448] In one embodiment, a base station may not infer at least one channel through an artificial intelligence model when it receives activation information from a terminal indicating not to infer a channel.

[0449] In one embodiment, a base station can infer at least one channel from a time point at which it receives activation information from a terminal instructing it to infer at least one channel until a time point at which it receives activation information instructing the base station not to infer at least one channel.

[0450] According to one embodiment, a base station may receive trigger information from a terminal via UCI instructing the terminal to infer at least one channel, if the pattern information indicates that the period of at least one SRS for inferring at least one channel is aperiodic.

[0451] According to one embodiment, when a base station receives trigger information from a terminal, the base station can infer at least one channel through an artificial intelligence model based on the trigger information.

[0452] At step 1440, the base station can determine the deterioration of the artificial intelligence model.

[0453] Degradation can mean that the performance of an AI model is degraded.

[0454] According to one embodiment, a base station may determine degradation of an artificial intelligence model through at least one of SGCS, SINR, MSE, NMSE, MAE, or NMAE based on information about at least one channel received from a terminal, but is not limited thereto.

[0455] The terminal can update the artificial intelligence model based on the degradation.

[0456] Updating an AI model may include, but is not limited to, at least one of replacing a degraded AI model with a new type of AI model or re-training the degraded AI model.

[0457] A new type of artificial intelligence model may mean an artificial intelligence model that has not deteriorated among at least one artificial intelligence model stored in a base station according to one embodiment.

[0458] A new type of artificial intelligence model may mean at least one artificial intelligence model received by a base station from a server according to one embodiment.

[0459] According to one embodiment, a base station can retrain an artificial intelligence model in which deterioration has occurred by inputting data previously received from a terminal into an input terminal of an artificial intelligence model in which deterioration has occurred.

[0460] In one embodiment, a base station can retrain an artificial intelligence model that has experienced degradation based on data previously received from a terminal and output corresponding to data previously received from the terminal.

[0461] The data received from the terminal may include, but is not limited to, information about at least one of the number of slots and / or the spacing between adjacent slots or the number of resource blocks and / or the spacing between adjacent resource blocks.

[0462] The operation of the terminal described in FIG. 14 can be performed by the processor and transceiver included in the terminal described above in FIG. 4.

[0463] The operation of the base station described in FIG. 14 can be performed by the processor and transceiver included in the base station described above in FIG. 7.

[0464] A terminal according to one embodiment may include a transceiver and at least one processor connected to the transceiver.

[0465] The at least one processor may transmit channel inference capability information including information about an artificial intelligence model used for channel inference of the terminal to the base station.

[0466] The at least one processor may receive configuration information about at least one CSI-RS used for channel inference from the base station based on the channel inference capability information.

[0467] The at least one processor can infer at least one channel through the artificial intelligence model based on the setting information.

[0468] The at least one processor can determine degradation of the artificial intelligence model based on the at least one inferred channel.

[0469] The at least one processor can receive the at least one CSI-RS from the base station.

[0470] The at least one processor can measure at least one channel based on the at least one CSI-RS.

[0471] The at least one processor may determine degradation of the artificial intelligence model based on comparing the at least one measured channel and the at least one inferred channel.

[0472] The at least one processor may transmit the channel inference capability information to the base station based on at least one of uplink control information (UCI), medium access control-control element (MAC-CE), or radio resource control signaling (RRC signaling).

[0473] The at least one processor may receive configuration information regarding the at least one CSI-RS from the base station based on at least one of downlink control information (DCI), MAC-CE, or RRC signaling.

[0474] Channel inference capability information according to one embodiment may include at least one of an identifier of the artificial intelligence model, a size of the artificial intelligence model, a complexity of the artificial intelligence model, information about an input of the artificial intelligence model, or information about an output result of the artificial intelligence model.

[0475] The configuration information regarding at least one CSI-RS according to one embodiment may include at least one of pattern information of the at least one CSI-RS for inferring the at least one channel or location information of the at least one CSI-RS for inferring the at least one channel.

[0476] The at least one processor may receive activation information from the base station for determining whether to infer the at least one channel through at least one of MAC-CE or DCI, if the pattern information indicates that the period of at least one CSI-RS for inferring the at least one channel is semi-persistent.

[0477] The at least one processor may receive trigger information from the base station via DCI to infer the at least one channel, if the pattern information indicates that the period of at least one CSI-RS for inferring the at least one channel is aperiodic.

[0478] The at least one processor may update the artificial intelligence model based on the degradation.

[0479] The at least one processor may transmit channel inference capability information for the updated artificial intelligence model to the base station.

[0480] The at least one processor may receive configuration information regarding at least one CSI-RS used for channel inference of the updated artificial intelligence model from the base station.

[0481] A base station according to one embodiment may include a transceiver and at least one processor connected to the transceiver.

[0482] The at least one processor may receive channel inference capability information from the terminal, including information about an artificial intelligence model used for channel inference of the terminal.

[0483] The at least one processor may transmit configuration information regarding at least one CSI-RS used for channel inference based on the channel inference capability information.

[0484] The at least one processor may transmit configuration information regarding the at least one CSI-RS to the terminal based on at least one of DCI (downlink control information), MAC-CE, or RRC signaling.

[0485] A method of operating a terminal according to one embodiment may include a step of transmitting channel inference capability information including information about an artificial intelligence model used for channel inference of the terminal to a base station.

[0486] A method of operating a terminal according to one embodiment may include a step of receiving, from the base station, configuration information regarding at least one CSI-RS used for channel inference based on the channel inference capability information.

[0487] A method of operating a terminal according to one embodiment may include a step of inferring at least one channel through the artificial intelligence model based on the setting information.

[0488] A method of operating a terminal according to one embodiment may include a step of determining decay of the artificial intelligence model based on at least one inferred channel.

[0489] A method of operating a terminal according to one embodiment may include a step of receiving at least one CSI-RS from the base station.

[0490] A method of operating a terminal according to one embodiment may include a step of measuring at least one channel based on the at least one CSI-RS.

[0491] A method of operating a terminal according to one embodiment may include a step of determining deterioration of the artificial intelligence model based on comparing at least one measured channel and at least one inferred channel.

[0492] A method of operating a terminal according to one embodiment may include a step of transmitting the channel inference capability information to the base station based on at least one of uplink control information (UCI), medium access control-control element (MAC-CE), or radio resource control signaling (RRC signaling).

[0493] A method of operating a terminal according to one embodiment may include receiving configuration information regarding at least one CSI-RS from the base station based on at least one of downlink control information (DCI), MAC-CE, or RRC signaling.

[0494] Channel inference capability information according to one embodiment may include at least one of an identifier of the artificial intelligence model, a size of the artificial intelligence model, a complexity of the artificial intelligence model, information about an input of the artificial intelligence model, or information about an output result of the artificial intelligence model.

[0495] The configuration information regarding at least one CSI-RS according to one embodiment may include at least one of pattern information of the at least one CSI-RS for inferring the at least one channel or location information of the at least one CSI-RS for inferring the at least one channel.

[0496] A method of operating a terminal according to one embodiment may include receiving activation information for determining whether to infer the at least one channel from the base station through at least one of MAC-CE or DCI, when the pattern information indicates that the period of at least one CSI-RS for inferring the at least one channel is semi-persistent.

[0497] A method of operating a terminal according to one embodiment may include receiving trigger information from the base station through DCI, which instructs to infer the at least one channel, when the pattern information indicates that the period of at least one CSI-RS for inferring the at least one channel is aperiodic.

[0498] A method of operating a terminal according to one embodiment may include a step of updating the artificial intelligence model based on the deterioration.

[0499] A method of operating a terminal according to one embodiment may include a step of transmitting channel inference capability information for the updated artificial intelligence model to the base station.

[0500] A method of operating a terminal according to one embodiment may include a step of receiving configuration information regarding at least one CSI-RS used for channel inference of the updated artificial intelligence model from the base station.

[0501] A method of operating a base station according to one embodiment may include a step of receiving channel inference capability information including information about an artificial intelligence model used for channel inference of the terminal from a terminal.

[0502] A method of operating a base station according to one embodiment may include a step of transmitting, to the terminal, configuration information regarding at least one CSI-RS used for channel inference based on the channel inference capability information.

[0503] A method of operating a base station according to one embodiment may include a step of transmitting configuration information regarding at least one CSI-RS to a terminal based on at least one of downlink control information (DCI), MAC-CE, or RRC signaling.

[0504] An operating method of at least one of a terminal according to an embodiment of the present disclosure or a base station according to an embodiment may be implemented in the form of program commands that can be executed by various computer means and recorded on a computer-readable medium. The computer-readable medium may include program commands, data files, data structures, etc., alone or in combination. The program commands recorded on the medium may be those specially designed and configured for the present invention or may be those known and available to those skilled in the art of computer software. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tapes, optical media such as CD-ROMs and DVDs, magneto-optical media such as floptical disks, and hardware devices specially configured to store and execute program commands such as ROMs, RAMs, and flash memories. Examples of program commands include not only machine language codes generated by a compiler, but also high-level language codes that can be executed by a computer using an interpreter, etc.

[0505] Additionally, the operating method of at least one of the terminals or base stations according to the disclosed embodiments may be provided as a computer program product. The computer program product may be traded as a product between a seller and a buyer.

[0506] A computer program product may include a software program and a computer-readable storage medium on which the software program is stored. For example, a computer program product may include a product in the form of a software program (e.g., a downloadable app) distributed electronically by an electronic device manufacturer or through an electronic marketplace (e.g., Google Play Store, App Store). For electronic distribution, at least a portion of the software program may be stored on a storage medium or temporarily created. In this case, the storage medium may be a storage medium of a manufacturer's server, an electronic marketplace server, or a relay server that temporarily stores the software program.

[0507] In a system comprising a server and a client device, the computer program product may include a storage medium of the server or a storage medium of the client device. Alternatively, if a third device (e.g., a smartphone) exists that is communicatively connected to the server or the client device, the computer program product may include a storage medium of the third device. Alternatively, the computer program product may include a software program itself that is transmitted from the server to the client device or the third device, or from the third device to the client device.

[0508] In this case, one of the server, the client device, and the third device may execute the computer program product to perform the method according to the disclosed embodiments. Alternatively, two or more of the server, the client device, and the third device may execute the computer program product to perform the method according to the disclosed embodiments in a distributed manner.

[0509] For example, a server (e.g., a cloud server or an artificial intelligence server, etc.) may execute a computer program product stored on the server, thereby controlling a client device in communication with the server to perform a method according to the disclosed embodiments.

[0510] Although the embodiments have been described in detail above, the scope of the present invention is not limited thereto, and various modifications and improvements made by those skilled in the art using the basic concept of the present invention defined in the following claims also fall within the scope of the present invention.

Claims

In a terminal (400) of a wireless communication system, the terminal (400) Transmitter and receiver (420); and At least one processor (410) connected to the above transmitter / receiver (420); At least one processor (410) above, Transmit channel inference capability information including information about an artificial intelligence model used for channel inference of the terminal to the base station, Based on the channel inference capability information, receiving configuration information about at least one CSI-RS (channel state information reference signal) used for channel inference from the base station, Inferring at least one channel through the artificial intelligence model based on the above setting information, A terminal that determines the decay of the artificial intelligence model based on at least one channel inferred above. In the first paragraph, the at least one processor (410) Receive at least one CSI-RS from the base station, Measure at least one channel based on at least one CSI-RS, A terminal that determines deterioration of the artificial intelligence model based on comparing at least one measured channel and at least one inferred channel. In the first paragraph, the at least one processor (410) A terminal that receives configuration information regarding at least one CSI-RS from the base station based on at least one of DCI (downlink control information), MAC-CE, or RRC signaling. In the first paragraph, The channel inference capability information includes at least one of the identifier of the artificial intelligence model, the size of the artificial intelligence model, the complexity of the artificial intelligence model, information about the input of the artificial intelligence model, or information about the output result of the artificial intelligence model. A terminal, wherein the configuration information regarding the at least one CSI-RS includes at least one of pattern information of the at least one CSI-RS for inferring the at least one channel or location information of the at least one CSI-RS for inferring the at least one channel. In the fourth paragraph, the at least one processor (410) A terminal that receives activation information for determining whether to infer the at least one channel from the base station through at least one of MAC-CE or DCI, when the pattern information indicates that the period of at least one CSI-RS for inferring the at least one channel is semi-persistent. In the fourth paragraph, the at least one processor (410) A terminal that receives trigger information from the base station through DCI, instructing to infer the at least one channel, when the pattern information indicates that the period of at least one CSI-RS for inferring the at least one channel is aperiodic. In the first paragraph, the at least one processor (410) Update the artificial intelligence model based on the above deterioration, Transmit channel inference capability information for the updated artificial intelligence model to the base station, A terminal that receives configuration information regarding at least one CSI-RS used for channel inference of the updated artificial intelligence model from the base station. In a base station (700) of a wireless communication system, the base station (700) Transmitter and receiver (720); and At least one processor (710) connected to the above transmitter / receiver (720); At least one processor (710) above, Receive channel inference capability information including information about an artificial intelligence model used for channel inference of the terminal from the terminal, Based on the above channel inference capability information, setting information regarding at least one CSI-RS (channel state information reference signal) used for the channel inference is transmitted to the terminal, At least one channel is inferred through the artificial intelligence model in the terminal based on the above setting information, A base station, wherein degradation of the artificial intelligence model is determined based on at least one channel inferred above. In a method of operation performed by a terminal in a wireless communication system, A step (S810) of transmitting channel inference capability information including information about an artificial intelligence model used for channel inference of the terminal to a base station; A step (S820) of receiving configuration information regarding at least one CSI-RS (channel state information reference signal) used for channel inference from the base station based on the channel inference capability information; A step (S830) of inferring at least one channel through the artificial intelligence model based on the above setting information; and An operating method comprising a step (S840) of determining decay of the artificial intelligence model based on at least one channel inferred above. In the 9th paragraph, the step (S840) of determining the deterioration of the artificial intelligence model is as follows: A step of receiving at least one CSI-RS from the base station (S842); A step (S844) of measuring at least one channel based on at least one CSI-RS; and An operating method comprising a step (S846) of determining deterioration of the artificial intelligence model based on comparing at least one measured channel and at least one inferred channel. In paragraph 9, The channel inference capability information includes at least one of the identifier of the artificial intelligence model, the size of the artificial intelligence model, the complexity of the artificial intelligence model, information about the input of the artificial intelligence model, or information about the output result of the artificial intelligence model. An operating method, wherein the configuration information regarding the at least one CSI-RS includes at least one of pattern information of the at least one CSI-RS for inferring the at least one channel or location information of the at least one CSI-RS for inferring the at least one channel. In the 11th paragraph, the operating method is: An operating method comprising: receiving activation information from the base station to determine whether to infer the at least one channel through at least one of MAC-CE or DCI, when the pattern information indicates that the period of at least one CSI-RS for inferring the at least one channel is semi-persistent; In the 11th paragraph, the operating method is: An operating method comprising: receiving trigger information from the base station via DCI, the trigger information indicating to infer the at least one channel, when the pattern information indicates that the period of at least one CSI-RS for inferring the at least one channel is aperiodic. In the 9th paragraph, the operating method is: A step of updating the artificial intelligence model based on the above deterioration (S850); Step (S860) of transmitting channel inference capability information for the updated artificial intelligence model to the base station; and An operating method further comprising a step (S870) of receiving configuration information regarding at least one CSI-RS used for channel inference of the updated artificial intelligence model from the base station. In a method of operation performed by a base station in a wireless communication system, A step (S1110) of receiving channel inference capability information including information about an artificial intelligence model used for channel inference of the terminal from the terminal; and A step (S1120) of transmitting configuration information regarding at least one CSI-RS (channel state information reference signal) used for channel inference to the terminal based on the channel inference capability information; At least one channel is inferred through the artificial intelligence model in the terminal based on the above setting information, An operating method in which deterioration of the artificial intelligence model is determined based on at least one channel inferred above.

Images